/VisualBag-Homogaphy

Implementation of Visual Bag of words on FashionMNIST data and homography estimation and mosaic creating

Primary LanguageMatlab

Homography Estimation and Visual Bag of Words

This project contains the following:

  1. Homography_ready_made/ - A folder to easily access the mosaic images and run them
  2. code/ - folder containing souce code and dataset.
  3. report.pdf - report of the assignment
  4. nearest_neighbours/ - folder consisting of some of the 200 visualizations of the patches (closest neighbours to the mean)
  5. This README

Download the fashion-mnist_test.csv and fashion-mnist_train.csv from Fashion MNIST data github page.

TASK 1 - Visual Bag of words

Run : go to code/ and type RunAll_2014CSB1034 in the matlab console.


TASK 2 - Mosaic

Run code: go to code/ and type the following : [cropped_mosaic] = ComputeHomography_2014CSB1034(img1, img2) where img1 is the image 1 and img2 is the image 2. Sample images are present in Homography_ready_made/ folder.

Now choose 4 points in img1 and hit enter, and similarly choose 4 points in img2 and hit enter. The mosaic image will be shown after the code runs.

Homography_ready_made In order to show some ready made output, I have made a ready-made code that wouldn't require you marking the points. To run the code type : ComputeHomographyReady in the MATLAB console and see the output.


nearest_neighbours/ - directory consists of the visualizaions of the patches that are the closest to the mean clusters.


References : [1] : http://www.cs.cornell.edu/courses/cs6670/2011sp/projects/p2/project2.html - Yosemite and the campus images are taken from this site.